DUMMER (Dumb Uncomplicated Match ModelER) aims to find distant relationships between genetic sequences (nucleotide or protein). It's similar to HMMER, but much simpler, and aspires to be better.
In more detail, it finds similar regions between sequences and "profiles". A profile is a set of position-specific letter, deletion, and insertion probabilities: typically made from a family of related sequences.
This is a proof-of-principle for the paper A simple way to find related sequences with position-specific probabilities.
Originally, DUMMER used HMMER-generated "profiles" as an input. However, HMMER defines insertion and deletion probabilities slightly differently than DUMMER, so they are suboptimal for its algorithms.
dmmbuild is supposed to change that. It's based on HMMER's hmmbuild, and has been adapted to build profiles for DUMMER. Essentially, as opposed to the original plan-7 profile Hidden Markov Model, we aim to build the following fig-4 HMM:
As the original HMMER project, we depend on Easel. You'll need to clone both the DUMMER and Easel repositories, configure Easel and build both, as follows:
% git clone https://github.com/Padraig20/dmmbuild
% cd dmmbuild
% git clone https://github.com/EddyRivasLab/easel
% cd easel
% autoconf
% ./configure
% cd ..
% makeAfterwards, you should be good to go! (If not, please feel free to reach out!)